Bower. — Studies in the Phytogeny of the Filicales. VI. 7 
where by fusion the number of strands may be reduced to four (iii) ; but 
afterwards the number increases again as the blade is approached (iv), while 
strands may be given off to supply the lateral wings (v). The origin 
of these is extra-marginal, for the actual margin of the vascular horseshoe 
faces inwards. The small strand in the wing seen in v is not formed 
as a direct branch from the petiolar strands, but is one of those small fibrils 
which form the ultimate reticulations of the network in the blade. In vi 
the petiolar strands are seen further subdivided, and ranged apparently 
in two strata, the one adaxial, the other abaxial. 
Nos. vii-xvii are from a different leaf, but they take up the details 
from the condition seen in vi. Each of the two strata is composed of four 
strands, while the gap ( + ) between the two marginal strands is still present. 
Small fusion-strands are then formed, which pass from the abaxial to the 
adaxial strands (viii-x), and meanwhile, by extra-marginal branching, 
strands are passed off laterally into the wings (x), which now assume 
considerable dimensions. After settling down again to two groups of four 
(xi), the strands begin to segregate into three groups, destined for the three 
lobes of the blade (xii-xv). Fusions occur in the lateral groups, so that 
in each of them the main supply is a strand which appears three-armed 
in transverse section (xiii-xvi). But still the strands of the central lobe 
remain distinct, the two adaxial still representing the margins of the original 
horseshoe, and the space between them ( -f ) having been open throughout 
the whole development. Finally, however, they fuse into a single strand, as 
in xvii, which represents one lobe in section. The origin of the vascular 
supply to the wings, and the sori which soon appear upon them, calls for no 
special description. 
It is thus seen that the whole arrangement is a modification of the 
fundamental horseshoe, which is spread out laterally and compressed in an 
antero-posterior direction, so that it apparently forms two strata. Fusions 
then appear between the members of the two strata, which have approached 
near to one another by reason of the compression of the horseshoe (viii-ix). 
Lastly, the flattened series divides up into the three groups which supply 
the lobes of the blade. Comparing with what is seen in Cheiropleuria , l the 
condition is distinctly more complex here, though in both the horseshoe is 
maintained with free margins. But in Platycerium 2 there are fusions across 
the adaxial face of the leaf, that is between the margins of the horseshoe 
and antero-posterior fusions as well. Thus it has departed farther than 
Leptochilus tricuspis from the primitive horseshoe. Accordingly, Lepto- 
chilus takes a middle position between the two structurally, as regards 
the petiole. 
Passing to the sterile blade, the venation corresponds to that of 
Dipteris and Cheiropleuria , and it is, like theirs, expanded in a single plane. 
1 Ann. of Bot., 1915, p. 506, Fig. 8. 2 l. c., p. 509, Fig. 10. 
